Every event we've mapped.
Amazon is in talks to sell its custom Trainium AI chips directly to third-party data centers, moving beyond internal AWS use. This positions Amazon as a direct competitor to Nvidia in the merchant silicon market. The move signals a new front in the AI chip wars with major implications across the semiconductor and data center value chain.
Super Micro Computer finalized a $7B financing package (equity + convertible preferred) to fund production of its $39B AI server order backlog. The capital injection accelerates procurement across the entire AI server supply chain despite ongoing governance and export control scrutiny. This unlocks near-term revenue flow to chip, power, cooling, and component suppliers.
Apollo and Blackstone have structured a $35 billion private credit facility to finance Anthropic's massive compute infrastructure expansion. The deal is one of the largest AI infrastructure financing events in history, channeling capital directly into data center buildout, GPU procurement, and memory systems. Micron and Nvidia are cited as key beneficiaries with notably different exposure profiles.
Nvidia has made a $2 billion strategic investment in Marvell Technology and deepened collaboration on optical interconnects and custom AI datacenter silicon. Jensen Huang publicly touted Marvell as a potential next trillion-dollar company. The deal signals Nvidia's intent to vertically integrate AI infrastructure beyond GPUs into networking and custom ASICs.
Cerebras Systems (CBRS) secured a multi-billion dollar compute deal with OpenAI and announced an AWS partnership, validating its wafer-scale AI chip architecture as a credible alternative to NVIDIA GPUs. The deals signal enterprise-grade commercial traction for Cerebras's purpose-built AI inference and training silicon. This reshapes the competitive landscape for AI accelerators and cloud-native AI infrastructure.
Amazon is investing several billion dollars to build a data center campus in Montgomery County, Missouri, expected to create 400+ full-time tech jobs. The expansion reflects continued hyperscaler infrastructure buildout driven by AI and cloud demand. This is a large-scale, multi-year capital deployment across construction, power, cooling, and networking supply chains.
AMD and Meta announced an expanded strategic partnership to deploy 6 gigawatts of AMD GPUs across Meta's AI infrastructure. This represents one of the largest single GPU commitments ever announced, signaling AMD's growing hyperscaler penetration against NVIDIA. The deal spans AMD's Instinct accelerator line and will require massive power, cooling, and supply chain buildout.
Amazon secured a $17.5 billion loan as its AI infrastructure buildout drives a roughly $200B capital expenditure commitment, pressuring free cash flow. The move signals hyperscaler spending is accelerating despite debt market reliance. Prediction markets remain skeptical of an imminent AI capex bubble burst.
Microsoft announced Majorana 2, a quantum computing chip redesigned using AI, aiming for commercial usefulness by 2029. The chip is based on topological qubit architecture, a fundamentally different approach than competitors like IBM and Google. This signals a major long-term capital deployment into quantum hardware, materials, and supporting infrastructure.
AMD is positioned at the center of an accelerating server CPU and AI accelerator upgrade cycle, with Q2 FY26 server revenue growing 70% YoY. The MI450/Helios GPU roadmap and EPYC CPU share gains are driving a structural shift in data center silicon spend. Analysts argue the market is underpricing AMD's combined CPU+GPU platform opportunity as hyperscalers diversify away from single-vendor dependency.
IREN Limited signed a $1.6B purchase agreement with Dell for air-cooled NVIDIA Blackwell GPU systems, backing a previously announced $3.4B five-year managed AI cloud services contract. The deployment signals large-scale AI infrastructure buildout by a crypto-origin data center operator pivoting to AI cloud. Air-cooled Blackwell specifically narrows the supply and cooling chain.
Humanoid robotics startup Neura Robotics has secured up to $1.4 billion in funding led by Nvidia and Amazon, marking one of the largest bets on physical AI to date. The round signals that hyperscalers and chip giants view humanoid robots as a near-term commercial opportunity. Capital is flowing into the full stack: silicon, sensors, actuators, software, and manufacturing.
Oracle reported a 93% jump in cloud revenue and announced plans to spend $70 billion on AI infrastructure. The massive capex commitment signals Oracle's aggressive push into AI cloud services. Rising infrastructure costs and financing plans drew investor scrutiny alongside the strong results.
Amazon has secured a $17.5 billion syndicated bank loan specifically to fund continued AI infrastructure spending, following a recent bond sale. This signals accelerating capital deployment into data centers, GPUs, and supporting infrastructure. The move underscores hyperscaler conviction that AI capex will remain elevated through the near term.
OpenAI filed to go public at a reported $852 billion valuation — potentially crossing $1 trillion at pricing. Beyond the listing itself, an IPO converts the world's most-watched private AI bet into a public benchmark, and arms OpenAI with acquisition currency and cheaper capital for its compute buildout.
At WWDC 2026 Apple unveiled a Gemini-powered Siri built with Google, plus an AI Extensions system that lets users pick assistants like Claude as iPhone options. Apple is renting intelligence rather than building it — and redistributing the mobile AI market in the process.
Beijing unveiled a two-trillion-yuan (~$295B) five-year plan for sovereign AI infrastructure — domestic datacenters, chips, and models, anchored by China Mobile, China Telecom, and Huawei. It is industrial policy at export-control scale: a parallel AI stack built explicitly to route around US silicon.
NVIDIA announced its Vera CPU and RTX Spark 'superchip' for Windows AI PCs, a direct assault on Intel and AMD's last stronghold. The same Vera silicon anchors its Vera Rubin AI supercomputers — one architecture from datacenter to laptop.
SpaceX prices its IPO this week, targeting ~$75B in proceeds at a $1.75 trillion-plus valuation — the largest listing ever, trading on Nasdaq as SPCX. Starlink's cash engine plus Starship's option value go public, and an entire space-and-telecom value chain gets repriced around it.
